NANO LETTERS
Microsphere Templating as Means of Enhancing Surface Activity and Gas Sensitivity of CaCu3Ti4O12 Thin Films
2006 Vol. 6, No. 2 193-198
Il-Doo Kim,† Avner Rothschild,* Takeo Hyodo,‡ and Harry L. Tuller Department of Materials Science and Engineering, Massachusetts Institute of Technology, 77 Massachusetts AVenue, Cambridge, Massachusetts 02139 Received October 3, 2005; Revised Manuscript Received November 11, 2005
ABSTRACT Chemical and physical synthesis routes were combined to prepare macroporous CaCu3Ti4O12 thin films by pulsed laser deposition onto poly(methyl methacrylate) (PMMA) microsphere templated substrates. These films showed remarkably enhanced gas sensitivity compared with control films deposited on untreated substrates, demonstrating the virtues of combining thin film physical vapor deposition (PVD) techniques in concert with colloidal templates to produce macroporous structures of inorganic films with enhanced surface activity for applications in chemical sensors, catalysts, and fuel cells.
The pursuit of new materials with novel functionalities has led, over the past several years, to the use of colloidal templates as self-assembled building blocks for the fabrication of quasi-ordered submicrometer structures of various materials.1-4 This approach has attracted a great deal of attention especially for the fabrication of photonic crystals but also as a means of synthesizing advanced catalysts and chemical sensors. It has been difficult, however, to employ this method to synthesize complex compounds and to ensure high reproducibly and throughput, largely due to the nature of the wet chemistry routes used to infiltrate the voids between the colloids with the precursors.5,6 In this Letter we demonstrate successful application of microsphere templating to the fabrication of macroporous CaCu3Ti4O12 (CCTO) films via pulsed laser deposition (PLD) on poly(methyl methacrylate) (PMMA) templated substrates. PLD is a versatile thin film deposition technique, well-known for its ability to transfer complex compositions of different classes of materials, such as high-temperature superconductors and ferroelectric materials, from the target onto the substrate.7 As demonstrated in this work, the combination of chemical and physical synthesis routes benefits from the unique advantages of the respective methods, i.e., the self-assembly feature provided by the PMMA microspheres enabling production of quasi-ordered templates with submicrometer dimensions, and the deposition of thin films and multilayer structures * Corresponding author. E-mail:
[email protected]. † Current address: Optoelectronic Materials Research Center, Korea Institute of Science and Technology, PO Box 131, Cheongryang, Seoul, Republic of Korea. ‡ Current address: Department of Materials Science and Engineering, Faculty of Engineering, Nagasaki University, 1-14 Bunkyo-machi, Nagasaki 852-8521, Japan. 10.1021/nl051965p CCC: $33.50 Published on Web 01/21/2006
© 2006 American Chemical Society
with precisely and reproducibly controlled properties via physical vapor deposition (PVD). Upon subsequent removal of the organic template by thermal decomposition, the residual PVD deposited inorganic film exhibits a macroporous structure with large surface area and surface-to-volume ratio. Consequently, its catalytic activity and gas sensitivity can be expected to be markedly improved with respect to similar films deposited on untreated substrates as demonstrated in the following. The material studied in this work, CaCu3Ti4O12 (CCTO), has attracted much attention due to its extraordinarily high dielectric constant.8,9 Permittivity values as high as ∼105 for single crystals and ∼2.8 × 105 for polycrystalline ceramics have been reported,10,11 very unusual for a material with no evidence of ferroelectricity. While the earliest reports ascribed this unusual phenomenon to highly polarizable relaxation modes related to the crystal structure of CCTO (intrinsic mechanism),8-10 most researchers now attribute it to microstructural features (extrinsic mechanism) such as grain boundaries in polycrystalline specimens11,12 or twin boundaries and other planar defects in single crystals.13,14 In particular, the internal barrier layer capacitance (IBLC) model has been put forward as the most plausible model to explain the high permittivity of CCTO.11,12 Indeed, a recent report on the highly nonlinear I-V characteristics of CCTO ceramics and the observation that the breakdown voltage is proportional to the number of grain boundaries intersecting the current conduction path15 support the IBLC model.16 Such a phenomenon is well-known to promote gas sensitivity in semiconducting metal oxides such as ZnO, SnO2, and WO3, wherein variations in the grain boundary barrier height
induced by interactions with reactive gases such as O3, NO2, H2, and C2H4 lead to significant changes in the I-V characteristics.17-19 This suggests that polycrystalline CCTO should serve as an excellent candidate as a sensor material with high sensitivity. In this contribution, we report, for the first time, on the gas-sensing properties of CCTO and compare the relative response of thin films deposited onto microsphere templated substrates versus untreated substrates. The microsphere templating method was pioneered by Scott et al.20 as a means of systematically investigating the effect of microstructure on the properties of SnO2 gas sensors. Quasi-ordered layers (∼200 µm thick), comprising closely packed SnO2 microspheres (opaline structures) or nanocrystalline SnO2 walls surrounding air microspheres (inverse opaline structures), were fabricated by reverse and direct templating strategies, using silica and polystyrene microspheres, respectively.20 More recently, Hyodo et al.21 reported on the gas-sensing properties of inverse opaline SnO2 layers (∼20 µm thick) fabricated by direct templating strategy using PMMA microspheres with different diameters between 150 and 800 nm. In both Scott’s and Hyodo’s studies, wet chemical synthesis routes were applied using liquid precursors to infiltrate the voids between the colloids.20,21 In this study, a different approach was taken in which a PVD technique (PLD) is employed to deposit thin films onto microsphere templates subsequently removed by thermal decomposition. A similar approach of combining PVD techniques with microsphere templates has been used previously by Luo et al. for the preparation of macroporous Si films.22 In this work we extend this approach to the fabrication of inorganic materials with complex compositions which, in many cases, provide unique functional properties. Furthermore, for the first time, a direct comparison between templated and untemplated (control) films is made, demonstrating remarkably improved gas sensitivity of the templated films. This suggests that this approach can be utilized to improve the performance of thin film gas sensors integrated on Si (or other) substrates by means of standard PVD techniques such as PLD or sputtering. These deposition methods typically lead to dense polycrystalline films wherein only the external free surface is exposed to interactions with the gas phase, and consequently the gas sensitivity is rather poor compared with porous thick films of compacted and partially sintered nanoparticles produced by conventional ceramic processing routes such as screen printing or spraying.23 As demonstrated in this work, this drawback can be overcome by using microsphere templates prior to film deposition, leading to enhanced surface area and reduced interface area between film and substrate, both considered highly advantageous for gas sensing. The scheme in Figure 1 illustrates the procedure used in this study to fabricate the macroporous CCTO films. An aqueous suspension (2 wt %) of 800 nm PMMA microspheres (Soken Chemical & Engineering Co., Tokyo, Japan), ultrasonically dispersed in deionized water, was dripped by pipet onto Al2O3 substrates with interdigitated Pt electrode 194
Figure 1. (a) Schematic diagram of the processing method used to fabricate macroporous CCTO films on Al2O3 substrates with interdigitated Pt electrode arrays. (b) Optical micrographs of gas sensor test devices (10 mm × 15 mm) with macroporous CCTO films.
arrays (200 µm Pt fingers spaced 200 µm apart), and dried overnight in a desiccator at room temperature. CCTO films were deposited by PLD onto the microsphere-templated substrates, as well as onto untreated substrates for comparison purposes, from a CaCu3Ti4O12 target prepared by a conventional solid-state reaction method using high-purity CaCO3 (99.99%), CuO (99.9%), and TiO2 (99.99%) powders (Sigma-Aldrich Co., Milwaukee, USA). The powders were mixed at the stoichiometric ratio, milled for 24 h, calcined at 900 °C for 11 h, and then pressed into pellets that were sintered in a tube furnace at 1100 °C in air for 24 h. PLD deposition was carried out at ambient temperature and oxygen pressure of 4 Pa. Laser ablation was carried out at a fluency of 2.5 J/cm2 and a repetition rate of 25 Hz using a KrF excimer source (λ ) 248 nm). Following the PLD deposition, the samples were calcined in air at 800 °C for 2 h to remove the PMMA microspheres. X-ray diffraction (XRD) and Scanning Electron Microscopy (SEM) were used to examine the phase composition and microstructure of the films, respectively. Examination of the specimens after different processing steps revealed that the dried PMMA suspension had four to five close packed multilayers. However, the CCTO films were deposited only on the top layer, and consequently, after the PMMA decomposed during the calcination step, the Nano Lett., Vol. 6, No. 2, 2006
Figure 2. SEM micrographs of (a) the surface and (b) crosssectional view of the macroporous CCTO films following calcination at 800 °C. The inset in (a) emphasizes the grain structure of the hemispheres while the inset in (b) shows the honeycomb structure when viewed from below after peeling the structure off of the substrate. The arrows in (b) mark the network of voids between film and substrate (see text).
resultant films consisted of a monolayer of hollow CCTO hemispheres with diameter of ∼800 nm, commensurate with the diameter of the PMMA microspheres, as shown in the SEM micrographs in Figure 2. Despite the collapse of the supporting template during the calcination step, no adhesion problems were observed with the film remaining intact even following an adhesive tape peel test. This is attributed to the strong chemical bonds between film and substrate formed during the calcination step at 800 °C. Reducing the thickness of the PMMA template from four to five multilayers to a monolayer by control of deposition and drying conditions is expected to improve adhesion even further. We note that no special efforts were taken in this study to optimize the uniformity and packing of the PMMA templates. While these parameters are important for some applications, e.g., photonic crystals, they are not crucial for the performance of gas sensors which are quite tolerant to structural inhomogeneities.20,24 Nevertheless, uniformity and order can be improved by careful control of the deposition and drying conditions of the microsphere suspension, as reported by other researchers.3,26-28 The XRD measurements indicated that after calcination the films had a single-phase polycrystalline CCTO structure (space group symmetry Im3).8 The wall thickness of the CCTO hemispheres was ∼80-90 nm, and the grain size varied between ∼60 and 110 nm, estimated from SEM Nano Lett., Vol. 6, No. 2, 2006
micrographs (cf. Figure 2). These parameters can be controlled by adjusting the PLD deposition conditions and postdeposition calcination conditions. The hemisphere diameter can be easily controlled by selecting microspheres of different sizes. PMMA microspheres between 100 nm and 1.5 µm are commercially available (Soken Chemical & Engineering Co., Tokyo, Japan). We note that size effects of the PMMA microspheres on the pore size distribution, specific surface area, gas response kinetics, sensitivity, and selectivity between different gases were systematically investigated by Hyodo et al. using 150, 250, 400, and 800 nm PMMA dispersions.21 A network of voids between the film and substrate, as seen in Figure 2, provides easy accesses of the gas phase to the internal surfaces of the hemispheres. These voids (“burst holes”) were formed, most likely, as a result of the PMMA decomposition during the calcination step. Consequently, the specific surface area is at least twice as large (top and bottom surface) and the surface area four time as large (top and bottom surface plus the geometric effect of the hemispheres) as in films deposited on planar untreated substrates. This unique morphology is expected to enhance their chemical activity and, at the same time, mitigate deleterious interfacial effects between film and substrate. Both effects are highly advantageous for gas sensors. On one hand, since the gas sensitivity is proportional to the specific surface area,28 one wants to reduce the critical dimensions of the sensing material (i.e., the grain size in the case of porous ceramic layers of partially sintered grains or the film thickness in the case of dense thin films)23 well below the 100 nm range. On the other hand, film/substrate interactions become increasingly important and the interfacial charge at the film/ substrate interface adversely interferes with the surface interactions between the sensing film and the gas phase29 at such critical film dimensions, which often counteract the beneficial effects of reducing the film thickness to nanoscale dimensions. Thus, a synthesis strategy that enhances the surface area while, at the same time, reduces the interface area between film and substrate provides a unique solution to this dilemma. The electrical properties of the macroporous CCTO films were examined using both dc I-V and ac impedance spectroscopy measurements. The I-V measurements were carried out at temperatures between 20 and 200 °C under vacuum conditions (∼10-5 Pa). Highly nonlinear I-V characteristics with strong thermal activation were observed, as shown in Figure 3a. Similar nonlinear behavior has been reported recently for polycrystalline CCTO ceramics by Chung et al.,15 who attributed the high nonlinearity to space charge barriers at the grain boundaries. To deconvolute the individual contributions of the grains and grain boundaries to the overall impedance of the CCTO films, ac impedance spectroscopy measurements were carried out (in air) as a function of frequency (1 Hz < f < 1 MHz), temperature (30 °C < T < 600 °C), and dc bias (0 < VDC < 6 V). A similar approach was taken recently by the authors to study grain boundary effects on the electrical properties of bulk CCTO ceramics, as reported in more detail elsewhere.30 195
Figure 3. (a) I-V characteristics at T ) 20 and 200 °C. (b) Impedance spectra as function of dc bias at T ) 405 °C. (c) Arrhenius plot of the grain boundary resistance (RGB) as a function of reciprocal temperature 1/T. The inset in (b) is a magnification of the impedance spectra at high frequencies.
The impedance spectra were modeled using an equivalent circuit with two parallel RC elements (R ) resistor, C ) capacitor), for the grains and grain boundaries, connected in series. This model captures the relatively small bulk resistance in series with blocking grain boundaries31 that give rise to the large dc (total) resistance of our specimens. Typical impedance spectra of the macroporous CCTO thin films, obtained at T ) 405 °C at dc bias levels ranging from 0 to 6 V, are shown in Figure 3b. Semicircular arcs with nonzero high-frequency intercepts on the Z′ axis were observed (see 196
inset in Figure 3b). These high-frequency intercepts correspond to the grain resistance, and the diameter of lowfrequency arc corresponds to the grain boundary resistance. The temperature dependence of the grain boundary resistance at zero dc bias (VDC ) 0) followed Arrhenius-type behavior with activation energy of 0.74 ( 0.01 eV, as shown in Figure 3c. A similar value (0.82 ( 0.02 eV) was obtained for bulk CCTO ceramic specimens, as reported elsewhere.30 The nature of the large grain boundary resistance was examined further by carrying out dc bias dependent impedance spectroscopy measurements. The nonzero high-frequency intercepts of the large semicircular arcs do not vary with the applied dc bias (see inset in Figure 3b), consistent with the assignment of this resistance to the ohmic resistance of the grains. In contrast, their diameter was found to decrease with increasing dc bias (see Figure 3b), confirming that the grain boundary resistance is nonlinear. This behavior is indicative of back-to-back Schottky barriers at the grain boundaries,32 and based on the results in Figure 3c, a barrier height of the order of 0.74 eV is estimated. Since the grain boundary resistance dominates the overall resistance at low frequencies, this also explains the high nonlinearity in the I-V characteristics shown in Figure 3a. Thus, the highly nonlinear I-V characteristics and strongly activated nature of the leakage current point to the crucial role played by grain boundary barriers in controlling the electrical properties of the CCTO films under examination in this study. The sensitivity of the macroporous CCTO films toward H2, CO, and CH4 gases was tested at temperatures between 200 and 400 °C and compared with similar CCTO films deposited on untreated substrates. Both templated and untemplated (control) films were deposited, calcined, and tested under the same conditions. They were mounted on Al2O3 sample holders and contacted by Pt wires which were attached to the interdigitated electrode arrays on the Al2O3 substrates using silver paste (SPI Silver Paste Plus, SPI Supplies, Chester, PA). The sample holders were then inserted inside a quartz tube placed in a tube furnace. Pt/ Pt-Rh (type S) thermocouples were used to measure temperature in situ. The resistances of the CCTO films were measured in parallel during exposure to different gas compositions using bottled gases of dry air, 1000 ppm H2 in air, 1000 ppm CO in air, and 1% CH4 in air (BOC Gases, Riverton, NJ), together with mass flow controllers (Tylan UFC-1500A mass flow controllers and a Tylan RO-28 controller). Prior to the gas-sensing tests, the resistance was measured as a function of flow rate in dry air and was found to increase with increasing flow rate, a behavior indicative of electron transfer from an n-type semiconducting oxide to chemisorbed oxygen adions.33 Indeed, thermoelectric power measurements of CCTO ceramics have indicated that the majority carriers are electrons, rendering CCTO an n-type semiconductor.15 To eliminate interfering effects due to changes in gas flow rate, the gas-sensing tests were carried out at a constant flow rate of 200 sccm. As shown in Figure 4a, the macroporous CCTO films deposited on templated substrates were found to be quite sensitive to H2, whereas the sensitivity of the homogeneous Nano Lett., Vol. 6, No. 2, 2006
Figure 4. (a) The resistance response of CCTO films deposited on templated (blue curve) and plain substrates (red curve) during exposure to increasing concentrations (between 100 and 1000 ppm) of H2 in air. (b) The steady-state conductance of the macroporous CCTO film as a function of the H2 gas concentration. (c) The resistance response during cyclic exposure to 100 ppm H2 pulses. All measurements were carried out at T ) 300 °C under total gas flow rate (air + H2) of 200 sccm.
films deposited on untreated substrates was negligible. The remarkably enhanced sensitivity of the templated CCTO films is attributed to their macroporous structure (cf. Figure 2) with the high surface area promoting interactions with the gas and the relatively small interfacial area mitigating the deleterious effects between film and substrate, as discussed above. Nano Lett., Vol. 6, No. 2, 2006
Figure 4b shows that the magnitude of the steady-state conductance (G ) 1/R) of the macroporous CCTO films follows a power law dependence on the H2 gas concentration ([H2]), Ggas ) Gair + R([Η2])β, with a power exponent (β) of 0.45 ( 0.01. This is close to the ideal value of 0.5 that can be derived from surface interactions between chemisorbed oxygen adions and reducing gases such as H2 as in this case.23,33 The small deviation from the rational fraction value of 0.5 probably relates to either agglomeration or zones in the structure that are less sensitive to H2 than others, as modeled recently by Scott et al.20,24 One may expect that with more uniform and ordered templates, obtained by careful control of the microsphere deposition and drying conditions, the sensitivity could, perhaps, be increased toward the ideal value of 0.5 for perfectly ordered structures.23,33 By extrapolating the magnitude of the response to lower H2 concentrations and assuming a minimum detectable change of 5% (corresponding to 3σ, with σ being the average noise level in our measurements) in the relative conductance Ggas/Gair, the limit of detection is estimated to be ∼55 ppm at T ) 300 °C. This value is comparable to the detection limit of commercial Taguchi type H2 sensors. Note, in contrast to commercial sensors that have been optimized by adding special catalysts and optimizing their microstructure and operation conditions, no such optimization efforts have taken place in the case of the CCTO films and therefore it is expected that their sensitivity could be further enhanced by employing similar efforts. The response to H2 was also reversible, as demonstrated in Figure 4c. Significant and reversible response was observed even at temperatures as low as 200 °C. Below that temperature, the resistance was too high (>100 MΩ) to be measured with our experimental apparatus. However, based on the trends observed in the temperature range 200-400 °C, the CCTO films are expected to exhibit even higher sensitivities to H2 at lower temperatures (
